March 9, 2016
Bank of Japan
Research and Statistics Department

The Bank of Japan has revised the seasonally adjusted Money Stock Data.
Having obtained the final figures for the whole year 2015, we have selected the most suitable ARIMA models applied for individual series.

Details of the seasonal adjustment method are as follows.

Seasonal Adjustment for Money Stock Data

The models adopted for each indicator are shown below:

1)

The Bureau of Census X-12-ARIMA method, version 0.3 is used.

2)

The multiplicative model and logarithmic transformation are adopted for every series. The critical values for additive outliers and temporary changes depend on the length of the time-series data. Both the former series data (from the beginning of series to March 2003)* and the current series data (from April 2003 to December 2015) are used for seasonal adjustment. For the months after December 2015, seasonally adjusted figures are calculated by applying the predicted seasonal factors.

1.	The model of L (average amounts outstanding) has been revised.
2.	M1 (amounts outstanding at end of period) is seasonally adjusted by using data (in which the former series are linked to the current series) of the last 56 years. Please note that the X-12-ARIMA program can handle up to 720 observations, among which 48 observations are used for the projection of seasonal factors and 672 observations (56 years for monthly data) are used for seasonal adjustment.

Inquiries

Financial Statistics Group, Economic Statistics Division, Research and Statistics Department

E-mail : post.rsd5@boj.or.jp